Re: [PATCH] Don't enable gdbtk in testsuite

[Pedro Alves <palves@redhat.com>]

On 12/05/2014 06:55 AM, Yao Qi wrote:
> Pedro Alves <palves@redhat.com> writes:
> 
>> That patch removed it from the git repo, mirroring how CVS modules
>> worked.  In CVS, if you checkout the "gdb" module, you don't get
>> the gdbtk dirs, but if you checkout the insight module instead, you
>> get everything gdb, plus the insight bits: src/gdb/gdbtk subdir,
>> src/gdb/testsuite/gdb.gdbtk/, and maybe other bits.
> 
> Hi Pedro,
> I looked at insight and the date of the last commit in
> gdb/testsuite/ChangeLog is 2013-10-21.  Looks insight stops updating gdb
> after gdb migrates to git so I think it should be safe to remove
> testsuite gdbtk from gdb head.

insight's official repo is kind of stuck in a limbo.  CVS is dead,
so that's not where you should be looking.  There's no official git
repo yet, but AIUI, people are using this non-official repo instead:

  https://github.com/monnerat/insight

In any case, I don't see why we should treat insight bits in
gdb/testsuite/configure.ac any different from bits in gdb/configure.ac.
It's exactly the same issue.

>> So removing the testsuite support for gdbtk doesn't seem like
>> the right thing to do.  Particularly since we still have the
>> gdbtk bits in gdb/configure.ac.  IOW, I don't see how
>> src/gdb/testsuite/gdb.gdbtk/ not being around is different
>> from src/gdb/gdbtk/ not being around.  We should either keep
>> all support for gdbtk, or remove all of it.
> 
> It is aggressive to remove gdbtk bits from gdb/configure.ac, although
> there were some "insight end-of-life" discussions on insight mail list.

That topic was raised, but from those discussions it seemed clear
to me that there is still interest in it.  E.g.,

 https://sourceware.org/ml/insight/2014-q4/msg00007.html
 https://sourceware.org/ml/insight/2014-q3/msg00010.html

Personally, I wouldn't be adverse to importing gdbtk into the official
gdb repo.  I don't see how that could hurt -- we tend to worry
about insight anyway when we touch code that might affect it, like
most the deprecated_foo hooks.

> I am OK to revert my patch.

I think we should do that.

Thanks,
Pedro Alves